NBER WORKING PAPER SERIES
ASSETS WITH "WARTS":HOW RELIABLE IS THE MARKET FOR TECHNOLOGY?
Vincenzo PalermoMatthew J. HigginsMarco Ceccagnoli
Working Paper 21103http://www.nber.org/papers/w21103
NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts Avenue
Cambridge, MA 02138April 2015
We thank Rajshree Agarwal, Dan Breznitz, Chris Forman, Daniel Spulber, Marie and Jerry Thursby,Arvids Ziedonis, seminar participants at Georgia Tech, CCC doctoral consortium (MIT), 2014 Academyof Management annual meeting, for helpful comments and suggestions. We are indebted to DeloitteReCap for access to their data. We also thank IMS Health Incorporated for their generous supportand access to their data. The statements, findings, conclusions, views, and opinions contained andexpressed herein are not necessarily those of IMS Health Incorporated or any of its affiliated or subsidiaryentities. The statements, findings, conclusions, views, and opinions contained and expressed in thisarticle are based in part on data obtained under license from the following IMS Health Incorporatedor affiliate information service(s): IMS MIDAS™, June 1997 to December 2008, IMSA Health Incorporatedor its affiliates. Higgins acknowledges funding from the Imlay Professorship and Palermo acknowledgesfunding from the Kauffman Dissertation Program Fellowship. The usual disclaimers apply. The viewsexpressed herein are those of the authors and do not necessarily reflect the views of the National Bureauof Economic Research.
At least one co-author has disclosed a financial relationship of potential relevance for this research.Further information is available online at http://www.nber.org/papers/w21103.ack
NBER working papers are circulated for discussion and comment purposes. They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies officialNBER publications.
© 2015 by Vincenzo Palermo, Matthew J. Higgins, and Marco Ceccagnoli. All rights reserved. Shortsections of text, not to exceed two paragraphs, may be quoted without explicit permission providedthat full credit, including © notice, is given to the source.
Assets with "Warts": How Reliable is the Market for Technology?Vincenzo Palermo, Matthew J. Higgins, and Marco CeccagnoliNBER Working Paper No. 21103April 2015JEL No. L10,L24,L65,O32
ABSTRACT
Existing research has focused on why and when firms may choose to access the external technologymarket. Surprisingly, however, less is known about the reliability of the patents attached to these externaltechnologies in the face of litigation. “Weak” external patents expose a firm to the potential loss ofdownstream revenues. To address this question we construct a novel dataset of patent litigation inthe pharmaceutical industry. We exploit a change in U.S. patent law as a natural experiment to testwhether external patents are more reliable than those developed internally. We find that acquired patentsare more likely to fall during litigation; they are less reliable then internal technologies. Losses leadto an average reduction in market capitalization of $450 million. Overall, our results demonstrate thecritical importance of the underlying reliability of external patents and provides a cautionary note tothe potential benefits of accessing external technology markets.
Vincenzo PalermoMunk School of GLobal AffairsUniversity of Toronto1 Devonshire PlToronto, ON M5S [email protected]
Matthew J. HigginsScheller College of BusinessGeorgia Institute of Technology800 West Peachtree StreetAtlanta, GA 30308and [email protected]
Marco CeccagnoliScheller College of BusinessGeorgia Institute of Technology800 West Peachtree Street, NWAtlanta, GA [email protected]
1
1. Introduction
The development of new technologies has become a process that often involves many
firms in multiple stages (Arora et al., 2001). Expanding markets for technology allow firms to
boost their performance by combining different technological inputs acquired externally, such as
through licensing or acquisitions. While much research on this topic has focused on the supply-
side of the technology markets, recent work has begun to analyze the role of the demand-side of
the technology markets (Cassiman and Veugelers, 2006; Laursen and Salter, 2006), highlighting
the effect of externally acquired technologies on firm performance.
Patents and their enforcement strategies have also increasingly become a crucial
component of firm competitive advantage and patent enforcement has been identified as a
strategic capability (McGahan and Silverman, 2006; Somaya, 2012). Firms can exploit internally
or externally generated technology to boost productivity and firm performance, but once a
downstream product is commercialized its protection is limited by the uncertainty of patent
litigation (Lemley and Shapiro, 2005). Little is known, however, on the role of acquired patents
in case of litigation and their importance as a defensive strategy for protecting downstream
revenues. Specifically, if acquired patents are more likely to be considered invalid during
litigation, technology buyers can suffer from significant revenue losses. Therefore, it is
strategically important to understand the differences between acquired and internally developed
patents in the context of patent enforcement.
To fill this gap, we combine insights from the economics and law literature on patent
litigation (Lanjouw and Schankerman, 2001; Lemley and Shapiro, 2005; Somaya, 2012) with
research on markets for technology (Gambardella et al., 2007; Arora and Gambardella, 2010) to
provide new evidence on the reliability of acquired patents in litigation cases. In particular, we
2
analyze litigation challenges and outcomes of internally developed or externally acquired
patents, including the impact of patent litigation on subsequent revenue loss if patents are held
invalid.
To answer our research question, we focus on the pharmaceutical industry and on the
early entry of generic products through a specific regulatory mechanism called a Paragraph IV
certification or challenge.1 In the U.S., new branded chemical-based drugs are granted data
exclusivity that runs in parallel with patent protection. After the expiration of data exclusivity,
generic manufacturers can challenge a branded product by filing a Paragraph IV challenge with
the FDA that usually results in litigation.2 This unique setting offers us the opportunity to
compare internally generated and externally acquired patents and their ability to successfully
defend their incumbent position through litigation. Our results suggest that external patents are
more likely to fall during patent litigation. This loss is significant because it allows for early
entry by generic manufacturers, negatively impacting a pharmaceutical firm’s future revenues.
We approximate this loss through the use of an event study and find that a litigation loss
corresponds to a real reduction in firm market capitalization of between $446m and $451m.
Our results provide a note of caution to the benefits associated with external technology
acquisition in boosting innovative performance. To our knowledge this is the first paper that
separates and analyzes the effects of internal and external patents on the probability of litigation,
final litigation outcomes and the resulting impact on firm market value. We contend that the
1 Interestingly, and more generally, sophisticated investors have been using Inter Partes Review (IPR) petitions to challenge patents they feel should be invalidated. Investors appear to be searching for questionable patents, filing an IPR petition and concurrently selling short the company’s stock. IPR petitions were created by Congress in 2011 to fight patent-trolls and are evaluated by a panel of three administrative judges. http://www.wsj.com/articles/hedge-fund-manager-kyle-bass-challenges-jazz-pharmaceuticals-patent-1428417408 2 The interested reader can see Voet (2013) for a complete discussion of the Paragraph IV challenge process.
3
reliability of external patents during ex post litigation is as important as the value of external
technology itself.
Our findings also call into question the due diligence process firms undertake prior to
licensing a patent. If firms knowingly are licensing technologies with known patent deficiencies
or “assets with warts” then our results suggest that firms may need to identify and license
potential technologies prior to their initial patenting. In this manner, the licensing
(pharmaceutical) firm would be able to direct (and fund) the writing of the focal patent. As we
will discuss, these firms have more experience writing patents and greater resources to commit to
the patenting process.
The paper is organized as follows: Section 2 reviews the current literature on external
technologies and patent litigation. Section 3 describes our industry setting. We discuss our data
and empirical findings in Sections 4 and 5, respectively. In Section 6, we quantify the economic
value of a litigation loss and we discuss the implications of our findings and conclude in Section
7.
2. The reliability of externally acquired patents
In the last few decades the importance of external technologies in boosting innovative
activity has grown dramatically and an extensive stream of literature has developed focusing on
the role played by technology commercialization, especially through licensing (Teece, 1986;
Arora et al., 2001; Gans et al., 2002; Chesbrough, 2003; Arora and Ceccagnoli, 2006;
Gambardella et al., 2007; Arora and Gambardella, 2010). In many industries, firms utilize and
build on external innovation to maintain their competitive market position, suggesting that
markets for technology are a key component of a firm’s innovative effort. For instance, Scherer
(2010) shows that a larger proportion of revenues for pharmaceutical companies are derived
4
from products that were discovered outside of the firm. Similarly, Ceccagnoli et al. (2010)
support this finding in their sample of new drugs introduced into the market; almost half of the
patents linked to new products were developed outside the firm.
Despite the positive effects related to external technologies, there are potential downsides
that may affect the ability to realize the payoffs generated by technology acquisition. First,
contractual uncertainties can undermine the supply of technologies. For example, uncertainty and
opportunistic behavior may increase transaction costs thereby reducing technology transfer
(Williamson, 1985). Second, the level of tacit knowledge may reduce the positive spillovers
generated by the transfer. In fact, when knowledge is difficult to transfer, the incentive to acquire
external technologies may be reduced (Kogut and Zander, 1993; Ceccagnoli and Jiang, 2013).
Previous analyses on optimal patent policy have usually assumed that there is no
uncertainty about the scope of patent protection (Gallini, 1984; Gallini and Winter, 1985).
However, subsequent perspectives recognize that patent protection is imperfect until it
successfully survives a challenge in court (Shapiro, 2003; Lemley and Shapiro, 2005). As
Lemley and Shapiro (2005) explain, the strength of patents is linked to the examination process
and, in general, the structure of patent review favors the approval of weak patents. As a result,
almost half of challenged patents are found invalid when litigated. For this reason, patents have
been defined as “probabilistic” since they do not confer an absolute right to exclude imitators but
they confer the right to try to exclude them through litigation (Lemley and Shapiro, 2005;
Hemphill and Sampat, 2011).
Existing research has extensively focused on either the role of external patents in
boosting a firm’s innovative effort or on the determinants and conditions that facilitate
technology transfer. However, there is no clear evidence of the reliability of acquired
5
technologies. Previous work comparing the quality of internally generated and externally
acquired technology has focused on assessing the importance of the “lemons problem” in
affecting technology trade. Asymmetric information between technology buyers and sellers
could lead to lower-quality technology transactions, which would result in higher quality
internally developed technology. While empirical results to date are scant and mixed, a recent
review of this literature suggests the information problems have been overemphasized and that,
especially in the context of licensing in the pharmaceutical industry, “…licensed compounds
appear to be drawn from the same distribution as the internally generated compounds of the
licensor” (Arora and Gambardella, 2010). While we do not directly contribute to this debate, our
work shifts the focus of the comparison downstream by comparing the quality of internal and
external technology from the point view of its legal strength rather than the technological quality
(as measured by the probability of successful development).
Our focus is on the extent to which firms are able to appropriate the returns generated by
external technologies once embedded in commercialized products. The argument behind this
logic is that holding a patent in the early stages, and conditional on successful development, it
may be easier to appropriate the economic rewards of the technology. If acquired technologies
are more reliable, from a legal standpoint, than those developed internally, firms may reduce the
litigation uncertainty and increase their ability to appropriate the monetary effects of innovation.
Patents acquired through the markets for technology may differ in quality than those
developed internally. Large pharmaceutical companies typically have an in-house patent
department that has a quality assurance program that emphasizes best practices in drafting and
prosecution of patents (Knowles and Higgins, 2011). These in-house attorneys have been trained
in potential pitfalls that could lead to invalidation and are kept closely up to date on the
6
company’s own litigation experiences and those of similar companies. Patent attorneys can
easily access research notebooks and data that can be discussed with the scientists. This fully
transparent landscape is valuable to determine whether the final granted patent can be relied
upon by the company to a hold a market (Knowles and Higgins, 2011). These large companies
also have the resources to engage the most experienced, high quality external patent attorneys, if
needed.
In contrast, suppliers of technology such as smaller research-intensive firms may not
have any experience with Hatch-Waxman generic litigation (i.e., Paragraph IV challenges) or
global litigation scenarios. Moreover, the licensor may only have limited research assets and
little commercial experience. Often these companies are resource constrained (Lerner et al.,
2003) and patents are drafted by outside counsel who themselves may have limited experience in
these issues (Knowles and Higgins, 2011).
We know from conversations with senior industry executives that externally acquired
patents go through a significant due diligence process. However does this selection process
ensure lower quality patents are simply not acquired? While this may be the goal, the process is
more nuanced and opaque. For example, during due diligence counsel for the licensee (large
pharmaceutical company) may not be allowed to inspect notebooks or raw data before giving a
binding offer. In some cases access to inventors may even be restricted during the negotiation
phase, or important contracts or documents withheld or heavily redacted (Knowles and Higgins,
2011). Compounding these issues is often the limited supply of licensable drug leads within
highly disaggregated markets. As such, pharmaceutical companies may knowingly license
‘assets with warts’ (Knowles and Higgins, 2011). The result of this complex, nuanced and often
7
opaque process is that internally generated patents may end up exceeding the average quality of
externally acquired patents.3
In sum, due to the offsetting effects on the legal quality of internal and external patents
highlighted above, answering the question of reliability of external patents is inherently
empirical in nature. In what follows, we therefore turn to our empirical framework.
3. The pharmaceutical industry and its regulatory environment
The pharmaceutical industry provides a natural setting for our analysis. Because of the
regulatory environment in the U.S., generic manufacturers can litigate the patents of a branded
drug before they expire thereby potentially undermining the branded companies incumbent
position. There is an extant literature on generic entry (Scott Morton, 2000; Reiffen and Ward,
2005; Grabowski and Kyle, 2007) along with an emerging literature relating this entry to
competition and innovation (Branstetter et al., 2011, 2014). An additional stream of research has
discussed patent challenges and their role in affecting the length of market protection
(Grabowski, 2004; Grabowski and Kyle, 2007; Hemphill and Sampat, 2011). In general, an
increasing number of drugs are being challenged and those with larger sales attract more
competitors (Scott Morton, 1999; Grabowski and Kyle, 2007; Hemphill and Sampat, 2011).
Berndt et al. (2007) finds similar results on the increasing rate of Paragraph IV challenges.
Despite their focus on the impact of authorized generics, which may reduce the incentive for
entry by new firms, the authors conclude that the number of Paragraph IV challenges remains
high. 3 It is critical to note that a weak or ‘bad’ patent does not mean the drug molecule or underlying technology is ‘bad’. In fact, from our field interviews we actually anticipate the opposite – a pharmaceutical company may license a weaker patent, the proverbial ‘asset with warts’, precisely because of the promise of the underlying molecule or technology. In this respect, our paper builds upon Arora et al. (2009). They focus on the differences between internally and externally developed drugs and show that external products are equally likely to succeed as internally developed molecules. However, that paper implicitly makes an assumption about the relative reliability of the internal and external patents. Equal success of clearing clinical trials and making it to market means less if the underlying patents are not reliable.
8
This last stream of literature suggests that generic challenges are driven by branded drug
sales and as a consequence of this increase in competition, branded companies face an increased
uncertainty over future revenues and reduction of R&D incentives. While results converge
towards the role played by sales as an incentive for generic entry, there is little evidence on the
role played by patents and their characteristics in the pre-entry decision and lawsuit outcome.4
To our knowledge, Hemphill and Sampat (2011) provide the first attempt to link litigation
initiated by generic manufacturers to patent characteristics. They find that conditional on sales
and drug characteristics, “weaker patents”, defined by citations and family size, are more likely
to face Paragraph IV challenges. In a follow up study, Hemphill and Sampat (2012) expand their
findings on Paragraph IV challenges confirming that patents that do not refer to the drug’s active
ingredient draw more challenges. They argue that these types of patents are often used by
branded companies to extend the effective market life of a drug, however the disproportionate
increase in challenges towards these patents limits the effectiveness of “evergreening”.
Branded drug protection was fundamentally changed with the passage of the Drug Price
Competition and Patent Term Restoration Act in 1984, informally known as the Hatch-Waxman
Act (HW). HW was introduced with the intent to establish a balance between incentivizing
innovative pharmaceutical research on the one hand and providing access to cheaper (generic)
drugs on the other hand. Under HW once a drug is approved, pharmaceutical firms are required
to list materially relevant patents in the FDA’s Orange Book.5 The FDA does not actively
regulate the patents that are listed and only these identified patents can be used to protect the
4 It should be noted that while product sales are important, we see a wide variance in the distribution of sales of those products that get challenged (Grabowski and Kyle, 2007). 5 The FDA Orange Book Preface states the following: “The patents that FDA regards as covered by the statutory provisions for submission of patent information are: patents that claim the active ingredient(s); drug product patents which include formulation/composition patents; use patents for a particular approved indication or method of using the product; and certain other patents as detailed on FDA Form 3542.” Available at: http://www.fda.gov/Drugs/DevelopmentApprovalProcess/ucm079068.htm
9
drug in case of litigation.6 Since the FDA has not precisely defined “materially relevant”
innovations, however, branded companies enjoy a certain level of freedom in selecting the
patents to list.
HW introduced “data exclusivity” for branded drugs in parallel to patent protection. Data
exclusivity is an exclusive marketing right granted upon approval and it runs concurrently with
patent protection. It protects the ownership of the underlying clinical trial data and prevents the
entry by generic manufacturers during that time period. It was intended to provide branded
products monopoly protection should underlying patent protection be limited. The majority of
chemical-based drugs receive 5 years of data exclusivity protection. Orphan drugs receive 7
years of protection while reformulations of an existing product receive only 3 additional years.
Firms can gain an additional 6 months of data exclusivity for the addition of a pediatric
indication. More recently, the GAIN Act provides an additional 5 years of data exclusivity for
select antibiotics.
Data exclusivity was balanced by a system that facilitated generic entry. Under this
system, the FDA can approve a new generic drug through an Abbreviated New Drug Application
(ANDA). To be approved, generic manufacturers only have to demonstrate that their product is
bioequivalent to a referenced NDA’s branded product (as opposed to running their own costly
clinical trials). While there are four ‘certifications’ that a generic manufacturer may claim in
order to enter the market, our focus is in the fourth certification, Paragraph IV.7 This is the only
6 This lack of oversight has led to criticisms of potential gaming of patent listings (Bulow, 2004) and evergreening or attaching new patents in the FDA Orange Book after approval by the FDA (Hemphill and Sampat, 2012) 7 Other certifications reflect a less competitive choice: under Paragraph I and II, patent protection has already expired therefore generic competitors can directly enter the market. Generic manufacturers apply for Paragraph III certifications when patent protection is still active, however the generic version of the drug can be commercialized only after patent protection has expired.
10
pathway that allows for the direct challenge of underlying branded patents prior to their
expiration.
The process related to Paragraph IV challenges is not linear and it is conditioned by
several strategic decisions. The process starts with the Paragraph IV challenge by a generic
manufacturer. Paragraph IV applicants are required to list and identify all the patents they want
to challenge. As part of the challenge, the ANDA filer has to notify the patent holder about the
challenge and submit a detailed statement of the legal basis by which its product does not
infringe on the listed patents or why the patents are invalid. A generic product may not infringe
on the patent if the manufacturer has discovered a way to produce a bioequivalent drug through a
different process, it has discovered a different structure of the same active ingredient, or it has
adopted a different delivery mechanism. Alternatively, a patent may be considered invalid if it
was wrongfully granted, it was anticipated by prior art, or if the invention was in public use for
more than one year before the USPTO application (Herman, 2011).
Once a branded patent holder (pharmaceutical firm) receives notice, they have 45 days to
initiate a suit. If an ANDA filer (generic manufacturer) wins the lawsuit, HW awards a 180-day
exclusivity right to the first Paragraph IV applicant. It has been estimated that the 180-day
exclusivity period generates potential revenues of approximately $60 million, more than
compensating the generic challenger for the cost of litigation (Higgins and Graham, 2009).
Figure 1 summarizes the Paragraph IV challenge process.
<Insert Figure 1 Here>
As described by (Knowles, 2010), the regulatory framework faced by pharmaceutical
firms can be modified both by new legislation and by the courts that apply them. A change in the
interpretation of patent law, for example, may alter the validity of patents (or vice versa). We
11
specifically focus on one such critical 2007 Supreme Court case, KSR International Co. v.
Teleflex Inc. This case affected whether a patent was considered non-obvious based on existing
prior art and involved the placing of a sensor on a fixed pivot point of the accelerator pedal of an
automobile.8 At first, the Court of Appeals for the Federal Circuit (CAFC) affirmed the validity
of the patent using the “teaching, suggestion, or motivation” test (TSM test). Under this test, a
patent is proved obvious only if there is some motivation or suggestion to combine prior art that
can be already found in the prior art or in the knowledge of a person with ordinary skill in the art.
However, the U.S. Supreme Court revised the CAFC decision and decided the TSM test
was too narrow and rigid. It concluded that an innovation could be obvious even if prior art did
not teach, suggest or motivate the innovation. The Supreme Court considered the Teleflex patent
as obvious and invalid and in doing so introduced a broader (and vaguer) definition of
obviousness. With this broader standard in place the obviousness of some branded product
patents became questionable. Thus, KSR should make it more likely that a generic firm initiates a
Paragraph IV challenge. A recent report by PwC (2013) appears to bear this out; generic
litigation jumped from 43 cases during 2001-2006 to 77 cases in the more recent post-KSR 2007-
2012 time period (PwC, 2013).
4. Data description and methodology
Our sample consists of all the new chemical entities approved by the FDA between 1995
and 2004 and their reported patents listed in the FDA Orange Book. Our analysis is limited to
drugs approved up to 2004 in order to allow all our drugs to have the opportunity to be targeted
8 U.S. Supreme Court Case No. 04-1350. http://www.supremecourt.gov/oral_arguments/argument_transcripts/04-1350.pdf
12
by a Paragraph IV challenge. Our final sample consists of 324 unique chemical-based drugs
approved by the FDA between 1995 and 2004, covered by 773 unique patents.9
We linked the drugs and patents collected from the FDA Orange Book to several data
sources. First, litigation data was gathered from The Paragraph Four Report. This data includes
information on the number of Paragraph IV applicants, the products and patents being
challenged, the status of pending court cases and the ultimate court decision. Next, we obtained
drug level data on sales and promotion expenditure from IMS MIDAS™. Patent approval dates,
number of claims, citations, and type were collected from Delphion™, IMS Patent Focus™ and
the United States Patent and Trademark Office (USPTO). Finally, stock market data was
obtained from the Center for Research in Security Prices (CRSP).
We identify whether a patent was acquired or internally developed through the
reassignment database of the USPTO. We compared the original patent assignee with the
company that commercialized the focal drug; if the two were different we defined that patent as
externally developed. Descriptive statistics and correlations are provided in Table 1 and 2,
respectively. All financial variables are converted into constant 2000 U.S. dollars and foreign
currencies are converted by using their respective average twelve-month exchange rate against
the U.S. dollar.
<Insert Table 1 and Table 2 here>
Our empirical approach relies on the adoption of a difference-in-difference (diff-in-diff)
specification (Angrist and Pischke, 2008). The diff-in-diff estimator identifies the effect of an
exogenous event on two groups of observations (e.g., firms and state). The simplest set up
includes two groups observed for two time periods. One of the groups is exposed to a treatment
9 We exclude biologic-based drugs in our sample because they are not covered by the same HW procedures as chemical-based drugs. In short, there is not (currently) an equivalent Paragraph IV mechanism for biologic-based drugs.
13
(exogenous shock) in the second period but not in the first period. In a non-parametric approach,
the average gain in the second (control) group is subtracted from the average gain in the first
(treatment) group. This removes biases in second period comparisons between the treatment and
control group that could be the result from omitted variables given differences between the two
groups and it reduces biases from comparisons over time in the treatment group that could be the
result of trends.
Practically, we exploit the 2007 KSR U.S. Supreme Court decision as an exogenous
temporal shock to estimate the diff-in-diff model. We consider externally acquired patents as the
treated group to be compared to the control group of internally developed patents. Our main
reported results are from a linear probability model. We also show the results are robust to logit
and probit formulations. This robustness confirms that there is typically little qualitative
divergence between the different specifications (Angrist and Pischke, 2008).
First, we analyze the probability of Paragraph IV related litigation. Our intent is to
understand whether the new legal environment (post-KSR) has an impact on the possibility of
challenge after controlling for other important factors, such as drug sales. Second, conditional on
being challenged, we estimate the diff-in-diff model on the lawsuit outcome to estimate the
implications of the change in patent law on patent litigation. Finally, we conduct an event study
on the firms whose drugs faced a Paragraph IV challenge in order to estimate the economic
impact of litigation outcomes.
We follow McWilliams and Siegel (1997) to compute cumulative abnormal returns
(CAR). First we estimate the market model using OLS over a period of 250 days prior to the
event. The estimation equation is the following:
14
(1)
where Rit is the return for firm i at time t and Rmt is the market return. The estimated OLS
parameters represent the stock’s “normal” return with respect to the market in a period prior to
the event. The abnormal return (AR) is defined as the return during a time span that includes the
event, in our case the generic challenge, minus the estimated return accounting only for the
market effect. In other words, the abnormal return is the forecast error between the “actual” and
the “normal” rate of returns. Empirically it is estimated as follow:
(2)
After estimating the abnormal returns for each firm i at time t, the CAR variable is computed
as the cumulative value of the standardized abnormal returns. The CAR equation is:
(3)
where ARit is defined by Equation 2, SDit is the abnormal return standard deviation and k
represents the cumulative number of days. We consider different event windows as a robustness
check for our analyses: we assume k equal to 3, 5 and 7 days. The event date is defined as t=0
and it represents the date of the litigation outcome.10
Finally, we multiple CARs by relevant firm market capitalization data. We argue that this
monetized value of the abnormal return represents the potential change in the discounted value of
future streams of revenues. For example, a negative outcome of a Paragraph IV challenge should
elicit a negative CAR. When multiplied by a firm’s market capitalization, the value should
10 No confounding events were identified over any of the event windows tested.
15
reflect the discounted value of lost future revenues due to early generic entry. On the other hand,
if branded patents survive a Paragraph IV challenge, generic manufacturers cannot enter the
market. In this case, we expect no abnormal market response because there will be no
unanticipated expected change in the level of market competition. In other words, a branded
company that wins a Paragraph IV challenge will be expected to continue to generate the already
anticipated future revenues.
4.1. Dependent variables
We rely on data from The Paragraph Four Report to create a dummy that equals one for a
drug that was challenged in a given year between 1999 and 2010, and zero otherwise. We are
able to identify a total of 264 unique Paragraph IV challenges. Among our drugs about 51%
experience at least one Paragraph IV challenge; Figure 2 maps the trend in challenges on a yearly
basis.
<Insert Figure 2 Here>
Prior to 2003 only 33 drugs were challenged. In the subsequent 5 years (2004-2008),
however, this number increased to 132, with 99 of these challenges occurring in the final two
years. The horizontal lines (Figure 2) represent the average number of challenges in three
different periods. It is easy to visualize the impact of policy changes on the number of Paragraph
IV challenges. Firstly, after the introduction of the Medicare Modernization Act in 2003, generic
manufacturers embraced Paragraph IV challenges as a viable strategy due to lower litigation
costs.11 Secondly, we observe another shift in the number of drugs challenged after 2008. This
11 The Medicare Modernization Act has limited the ability to “stack up” multiple 30-month periods of protection, a strategy known as evergreening. This changed forced pharmaceutical companies to make all their claims against a generic manufacturer in their initial lawsuit in response to a Paragraph IV challenge (Bulow, 2004).
16
increase is related to changes in patent law due to the KSR decision; the average number of
challenges increased by 87% compared to the previous period.
Our first dependent variable is a patent level dummy, Paragraph IV (patent level), that
equals one if the ANDA generic filer challenges a focal patent in a given year, and zero
otherwise. Out of the 773 unique patents, 356 of them were challenged at least once in our data.
As shown in Figure 3, the distribution of patents per number of Paragraph IV challenges is
skewed towards zero. About 17% of our patents (131 patents) received at least 2 challenges, and
29% (225 patents) were litigated only once. Among the 417 patents that were not litigated, 349
patents are listed for drugs that did not receive any Paragraph IV challenges while 68 were not
challenged but listed under drugs that were litigated.
<Insert Figure 3 Here>
Our second dependent variable classifies the Paragraph IV lawsuit outcome. We identify
two potential scenarios: first, the court decides on the validity of the litigated patents, and
second, the companies can privately settle the case. For the latter, we collected data from
companies’ statements and SEC filings to identify the terms of the agreements. We categorized
four different outcomes:
1. The court rules in favor of the Paragraph IV applicant (i.e., generic manufacturer). One or
more patents listed in the FDA Orange Book are considered either invalid or not
infringed on by the generic product. It follows that the generic manufacturer can enter the
market
2. The parties settled prior to trial and the agreement, in contrast to the previous case, allows
generic manufacturers to enter as an “authorized generic”. We coded this outcome as
17
favorable to generic manufacturers because the branded drugs market share will begin to
diminish.
3. The parties settled prior to trial, but the agreement either delays or blocks generic entry.
We coded this outcome as favorable to a pharmaceutical company.
4. The court rules in favor of the pharmaceutical company. Patents listed in the FDA
Orange Book are considered either valid or infringed by the generic product and
therefore, the generic manufacturer cannot enter the market.
To create the outcome dummy variable, we classify the first two cases as favorable to
generic manufacturers and the last two cases as favorable to pharmaceutical companies.
Therefore, Paragraph IV outcome is a dummy that equals one in case of favorable outcomes for
generic manufacturers, and zero otherwise. On average, a favorable outcome for generic
manufacturers occurs in about 47% of the cases and is almost always equally divided between
court decisions (63 occurrences) and settlement agreements (61 cases).
4.2. Independent variables
External Patent is a dummy that takes a value of one if a patent was originally assigned
to a different company than the pharmaceutical company that marketed the drug, and zero
otherwise. With the available data we are not able to identify the method of procurement (i.e.,
licensing or acquisition) because we can only track changes in patent assignees over time and
compare the original assignee with the company that marketed the branded drug. The number of
external patents does not significantly differ from internal patents; 383 patents in our sample
(about 49%) represent external technologies. Additionally, only 41% of these patents (159 out of
18
383 external patents) are challenged compared to 50% of the internal patents (197 out of 390
internal patents).
To implement our diff-in-diff estimation strategy, we create a dummy (2007 Dummy) that
equals one for all the observations after the 2007 KSR Supreme Court decision. The dummy
variable identifies the time period after the Supreme Court decision, which we consider an
exogenous event. While the case was known the final determination of the case was unknown as
was its impact on the pharmaceutical industry. Ultimately, the case represented an important
legal change that affected the outcome of Paragraph IV challenges.
Next, we define yearly branded product level sales, Drug Sales, as one of our drug-level
independent variables. Our intent is to study whether higher revenue drugs influence the decision
for entry by generic manufacturers since more profitable segments may attract more competitors.
Prior literature has shown that more profitable drugs have a higher probability of being attacked
(Grabowski and Kyle, 2007; Hemphill and Sampat, 2011). Similarly, Caves et al.
(1991)_ENREF_11 find that market share is a key determinant of generic entry.
To control for differences across types of patents we include a set of dummies, Type of
Patent. We rely on data from the IMS Patent Focus™ database that describes the function and
use of focal patents. Each patent is categorized into one of the following groups: product patent,
compound patent, method of use patent, drug delivery system and “other”, which includes
process patents. Based on this classification, we create five dichotomous variables. In our sample
approximately, 21% of the patents are products, 21% are classified as method of use patents,
34% of the patents protect the drug composition, and finally, 14% are drug delivery system
patents. On average, we do not find significant differences in the distribution of patent types
between internal and external technologies.
19
Based on prior research, we also include four variables to control for patent
characteristics (Allison and Lemley, 1998; Lanjouw and Schankerman, 2001; Lemley and
Shapiro, 2005). First, we collected data on both Forward and Backward citations. Forward
citations are measured by cumulating the number of citations received by a patent from its grant
year up to any given year and represent the impact of the focal patent on subsequent innovations.
Patents with more forward citations may be more “important” due to their impact on future
innovation. Additionally, the economic value of patents may be positively related to the number
of times the patent is cited. Backward citations, on the other hand, equal the number of existing
patents cited and denote how fundamental and innovativeness of a patent. Patents with
significant numbers of backward citations extensively build on existing knowledge and therefore
may be less innovative. Among the challenged patents in our sample, external technologies have
significantly fewer backward citations than those developed internally.
Second, we use the numbers of Claims as a measure of patent quality (Lanjouw and
Schankerman, 2004). The principal role of claims is to define and detail the novel features of the
invention. We construct a variable that counts the total number of claims in all the patents in our
sample. There are not any significant differences between internal and external patents.
Third, following Hemphill and Sampat (2011), we control for the positive effect that late
expiring patents can have on the market life to the focal drug and that the timing of patenting
may affect generic entry. New Patent dummy takes a value of one if, within the patent portfolio
for a single drug, the grant date of a patent is the latest, zero otherwise. By doing so we are able
to trace from a temporal point of view which patents for a specific drug have the latest grant
date. Particularly in the pharmaceutical industry, the timing of technology patenting is not
necessarily coincident with its commercialization.
20
Next, we add controls for drug specific characteristics. Patent per innovation controls for
the total number of patents attached to the focal NDA in the FDA’s Orange Book. By doing so,
we take into consideration the different sizes of the set of patents protecting each drug. On
average, drugs in our sample have four patents listed in the FDA Orange Book while the drug
with the largest set has 18 patents attached.
Companies may also have different incentives in protecting their products. In particular,
we control for the relative importance of a focal drug to a specific company. We define Drug
importance as drug sales divided by firm sales in any given year. We control for the drug’s
relative importance because we assume that in case of a Paragraph IV challenge a company will
be more likely to commit greater resources to their most valuable products.
Finally, we include a dummy to account for challenges that include Teva Pharmaceutical
Industries (Teva dummy). Teva is the largest generic company and has been active with
Paragraph IV challenges.12 In our sample Teva is involved with about 25% percent of Paragraph
IV cases (66 cases out of 264). We build a dichotomous variable that equals one if Teva is
among the first group of challengers in a Paragraph IV challenge, zero otherwise.
5. Results
5.1. Probability of receiving a Paragraph IV challenge
In our first set of analyses, we estimate the probability of a Paragraph IV challenge at the
patent level. Table 3 reports these estimates based on a variety of empirical specifications.
Models 1 through 5 report the results computed with a linear probability model, Model 6 through
10 are estimated with a probit model while Models 11 through 15 are based on a logit model. We
find no evidence of a difference between internal and external patents; the interaction terms 12 http://www.tevapharm.com/en-US/About/Pages/AboutUs.aspx. Access 09/26/2011
21
between our primary dummies (External Patents and 2007 Dummy) are never significant. These
results should not be surprising given the common practice to challenge almost all patents listed
in the FDA Orange Book.
<Insert Table 3 Here>
Consistent with prior findings, we find that drug sales drive most of the variation in
explaining the probability of a generic challenge (Scott Morton, 1999; Grabowski, 2004;
Grabowski and Kyle, 2007; Hemphill and Sampat, 2011). As demonstrated by Higgins and
Graham (2009), the first generic producer to enter the market is granted 180 days of market
exclusivity, which translates into approximately $60 million in profits. Therefore, there is more
of an incentive to attack and focus on the most profitable drugs. In addition, it has also been
shown that sales affect the intensity of challenges, conditional on a challenge occurring (Berndt
et al., 2007; Hemphill and Sampat, 2011).
We find a positive relationship between the 2007 KSR Supreme Court decision, 2007
Dummy, and an increase in litigation. This suggests that generic manufacturers are increasing
their Paragraph IV challenges in order to exploit the new post-KSR legal environment. Given the
less restrictive validity test for patents, generic manufacturers may perceive the new legal
environment as more favorable, thus creating more incentives to challenge branded drugs. This
increase in Paragraph IV challenges was an unintended consequence of KSR. We could find no
discussion of this feared development in the KSR decision or in any of the Amicus briefs filed in
the Supreme Court case.
5.2. Paragraph IV outcomes
22
After generic manufacturers file a Paragraph IV challenge, entrance into a market is
defined by either a positive outcome of the litigation or by a favorable settlement with the
pharmaceutical firm.13 Therefore, it is important to understand the role of patents in determining
the litigation outcome for each challenge. We do so by estimating litigation outcomes using a
diff-in-diff approach. Table 4 reports the estimated coefficient of this Paragraph IV outcome
regression. The dependent variable equals one if a generic manufacturer is able to enter the
market by either winning a challenge or by signing an agreement with a branded company to
become an authorized generic. The variable equals zero when the court rules in favor of branded
companies (i.e., litigated patents are valid and/or infringed) or when pharmaceutical companies
adopt a “pay-for-delay” strategy (Hemphill, 2006). We estimate our models through 3 different
specifications: linear probability models (Models 1 to 5), probit models (Models 6 to 10) and
logit models (Models 11 to 15).
<Insert Table 4 Here>
Across all specifications, the interaction between External Patents and 2007 Dummy is
always positive and significant. This result suggests that acquired patents are less reliable than
those drafted internally. More specifically, after 2007 (post-KSR), external patents are more
likely to be considered invalid thus allowing for increased generic entry via Paragraph IV
challenges. Our findings suggest that there may be patent validity risk associated with the
acquisition of external patents and it supports the IP validity risk described by Knowles and
Higgins (2011). Pharmaceutical firms may very well be acquiring ‘assets with warts’. The shift
13 Generic manufacturers can enter the market before court ruling. However, generic producers must weigh the benefits and risks of an “at-risk launch” to minimize possible downstream risk in case the court rules in favor of the pharmaceutical company.
23
in patent law as a result of KSR reveals that external patents, on the margin, were more sensitive
to the change in non-obviousness.
Over the last few decades the pharmaceutical industry has increasingly turned to the
external technology markets to obtain new drug candidates and novel technologies (Arora et al.,
2001; Chesbrough, 2003; Cassiman and Veugelers, 2006; Ceccagnoli et al., 2014). However, the
external technology market is not a homogenous field of patenting capabilities. Firms may very
well be acquiring patents from companies with limited patenting experience. Few small,
research-intensive firms have the resources to hire high quality IP firms. And while a firm may
receive a patent, the deficiencies in the process and construction may ultimately affect its
reliability during litigation.
On the demand side, pharmaceutical companies have extensive experience in the external
technology markets and conduct a due diligence process that should reduce the risk of acquiring
non-optimal patents (Arora et al., 2009). However, in practice the licensing decision is often
driven by a specific technology need to refill the R&D pipeline and whose supply may be very
limited. Such circumstances can create situations, as discussed earlier, where patent attorneys
may not have access to clinical notebooks or key scientific personnel during the due diligence
process (Knowles and Higgins, 2011). In extreme cases, potential licensors are required to make
a binding offer prior to having access to these items (Knowles and Higgins, 2011).
Consequentially, the decision to acquire external patents may rely on an imperfect set of
information and reduce the ability to evaluate the validity of a patent. In essence, pharmaceutical
firms may knowingly acquire ‘assets with warts’, a very different outcome than previously
assumed in the markets for technology literature.
24
Finally, while we are unable to control for other lawsuit variables as suggested by Allison
and Lemley (1998) we are able to proxy for the “willingness to protect” a drug by focusing on its
relative importance to overall firm sales. We assume that pharmaceutical firms would adopt a
more aggressive defensive strategy (and committing more resources) to those drugs that
represent high volume of sales or have benefited from significant sunk costs, such as detailing or
considerable downstream investments. As expected, our results confirm that pharmaceutical
companies tend to allocate more resources to the defense of those drugs that are more important
to the firm.
6. Economic Impact
Successful Paragraph IV challenges have detrimental economic consequences for
branded companies that lose their monopoly market position due to early generic entry. An
unanticipated loss of market position and resulting revenues should elicit a negative equity
response. We estimate the potential impact of successful Paragraph IV challenge (loss for the
branded pharmaceutical firm) with an event study at the time of the litigation decision.
Following McWilliams and Siegel (1997), we make the following assumptions. First, we
assume that markets are efficient. The efficient market hypothesis implies that capital markets
are effective in incorporating all relevant information available to traders into the stock price of a
firm. Under this assumption, equities should respond when new relevant information is revealed.
Given our empirical setting, we argue that the litigation outcome of a Paragraph IV challenge
represents precisely such an event.
Second, it is important that events are publicly announced so that markets may digest the
news. These types of suits are regularly reported in the general news as well as practitioner
publications and websites. Because of the probabilistic nature of patents (Lemley and Shapiro,
25
2005), it is rarely obvious which firm will prevail during court. In case of a pharmaceutical win,
markets will gain no new information from the announcement that would influence equity price.
The market position of the branded product will remain intact as will the flow of already
anticipated future revenues. If current equity price reflects discounted future cash flows, then
nothing from a pharmaceutical win will disturb this calculation. In the context of the event study
we would expect not to see any abnormal response since there is no unanticipated value being
created or destroyed.
On the other hand, in case of successful challenge by a generic manufacturer, early entry
of new competitors in the market will dramatically erode branded product sales. This loss in
future revenues should negatively impact equity price. In terms of our event study, this
unanticipated negative shock should be reflected in a negative CAR value. Moreover, the
monetized value of this abnormal response should reflect the value of the lost future revenue.
Table 5 reports CARs for two subsamples, cases where the pharmaceutical company wins
and cases where the pharmaceutical company loses, for three different event windows
encompassing the Paragraph IV litigation decision date. In contrast to Panattoni (2011), across
all event windows we find no significant CAR in the case where the pharmaceutical firm wins
the case. If we assume that markets are efficient then this result should not be surprising; there is
no reason to expect an abnormal market response. While a “win” might be unexpected, the
resulting market position of the branded drug will not change nor will the already anticipated
stream of future revenues, which should already be reflected in equity prices.
However, in case of loss, we find significant negative cumulative abnormal returns
suggesting that the legal outcome was an unanticipated negative shock. In these cases
pharmaceutical companies lose their incumbent position due to early generic entry, suffering
26
significant loss of future revenues. Across the three event windows, the CAR is significant for
our two longer event windows. As we have correctly accounted for any potential confounding
events, this difference suggests that there may be a leakage of information prior to the
announcement date.
<Insert Table 5 here>
In order to quantify the actual loss in firm value we collected firm’s market capitalization
for the same month and year of our events. For our sample of losses, we estimate the average
market capitalization and multiply this by the estimated CARs. For our two significant windows
we find losses between $446m and $451m, respectively. The decline in firm value, representing
the loss in future revenues, is significant for a pharmaceutical company when compared to the
average cost of developing a new approved drug, which in the early 2000s, was equal to $1.2B
(DiMasi et al., 2003). This negative response represents about 30% of the R&D cost necessary to
develop a new drug. In line with Branstetter et al. (2014)_ENREF_9, these results provide one
mechanism (i.e., decline in revenues) as to why early generic entry may reduce the incentive for
new drugs within a specific therapeutic market.
7. Discussion and conclusion
We use a unique patent litigation dataset to examine the relative reliability of external
patents. Our empirical context is the pharmaceutical industry and litigation arises from Paragraph
IV challenges filed by generic manufacturers. This paper expands our understanding of the
importance of acquired patents in protecting future downstream revenues. It is commonly
accepted that acquired technologies can increase innovative productivity, generate knowledge
27
spillovers and create unique synergies with existing internal competences (Arora et al., 2001;
Gans et al., 2002; Cassiman and Veugelers, 2006; Ceccagnoli et al., 2014). However the benefits
in terms of future revenue streams are not guaranteed from these activities. Instead, those
revenue streams will only be protected if the externally acquired patents are reliable during
litigation.
By relying on the markets for technology and patent litigation literatures, our study
provides novel insights on the role that acquired patents play during ex post litigation. In our
research setting we find evidence that external patents are less reliable than those developed
internally. In fact, our results suggest that external patents increase the possibility that generic
manufacturers are able to successfully win their Paragraph IV challenges. Assuming that the
market is correctly discounting lost future revenues, we estimate these losses at between $446m
and $451m.
It has been argued that pharmaceutical companies can effectively select patents available
in the market (Arora et al., 2009). However, the legal shift induced by the Supreme Court in KSR
(from the “TSM test” to a broadly defined “common sense” test) that we exploited in our
analysis suggests that firms may very well be selecting some ‘assets with warts’. On the margin,
external patents in the post-KSR era were more likely to fall during litigation, suggesting that
their reliability was more tenuous to begin with. Our findings suggest, contrary to prior belief,
that the due diligence process is not summarily looking past weaker patents.
Why would a pharmaceutical firm knowingly acquire ‘assets with warts’? Our field
interviews, consistent with prior work (Knowles & Higgins, 2011), seem to provide two
explanations. Firstly, pharmaceutical companies have the financial resources and experience to
craft higher quality patents, taking into consideration broader IP strategies. However, smaller,
28
research-intensive firms are often resource constrained and as a result patents are written by
attorneys with more limited experience or familiarity with pharmaceutical litigation, including
Paragraph IV challenges (Knowles and Higgins, 2011). Therefore, the reliability of these patents,
on average, during litigation may be lower.
Secondly, firms acquire external technologies to create synergies with internal
capabilities (Cassiman and Veugelers, 2006; Ceccagnoli et al., 2014) and to refill product
pipelines (Higgins and Rodriguez, 2006). At times the need of the pharmaceutical firm may be
great (and the supply of technology more limited) such that they knowingly take on this
reliability risk. More broadly, our results do suggest that if firms are going to acquire external
technologies they should do so prior to external patent construction so that the larger, acquiring
firm can put their considerable IP resources to work writing a more reliable patent.
There are other reasons that may lead to this outcome. For example, it is often the case
that a licensor will retain prosecution control while the licensee must defend the prosecution
decisions during litigation. The licensor may also make short-term motivated decisions in
prosecution to accomplish a quick patent grant, which can add to longer-term litigation risk
(Knowles and Higgins, 2011). Such a nuanced distribution of control rights calls for an explicit
analysis of licensing contracts, a more clear understanding of how rights are distributed and,
ultimately, their effect on litigation outcomes.
A natural question to ask, if our findings are correct, is whether this litigation risk is
accurately reflected in the price of the license. Our conversations with industry practitioners
seem to indicate that this risk is not fully appreciated. If this is the case, are firms, on average,
overpaying for external technologies that are already covered by patents? Another area for future
research would to explore a more complete valuation model that includes litigation risk.
29
Disaggregate data, at least for the pharmaceutical industry, exists such that it would be possible
to explore these issues.
Ultimately, this heretofore-underappreciated aspect of the external technology markets
has profound implications on future R&D expenditures and innovation. For example, some have
called for the movement away from a traditional R&D model to one of “S&D” or “search and
develop” (Morgan Stanley, 2010). That is, large pharmaceutical firms should focus their
capabilities on development (clinical trials) while at the same time acquiring all their drug
candidates. And while not this extreme, we do see some companies, such as Glaxosmithkline
allocating close to 50% of their R&D budget towards externally acquired technologies. Such
calls need to be dampened, we believe, and take into account not just the technologies (which
may be superb) but also their associated patents (which may be poor). Given that future
innovation and current R&D are paid for from existing product sales the importance of stable
and reliable future revenue streams cannot be understated. If internal productivity within the
industry has slowed and externally acquired patents may not reliable, the answer to the industry’s
overall productivity crisis becomes more complex.
Finally, recent work is beginning to document the longer-term impacts of generic
penetration into markets on pharmaceutical innovation (Branstetter et al., 2011). The authors are
agnostic about how generic entry occurs (whether via Paragraph III or Paragraph IV) but rather
focus on how quickly generics penetrate into a market. Our work complements that paper
because we explain what might be occurring behind the scenes via Paragraph IV challenges,
especially as it relates to externally acquired patents. They demonstrate that as a result of entry
pharmaceutical firms appear to be shifting away from chemical-based products towards
biologics-based products. The rotation is logical given that the current legislative framework
30
created under Hatch-Waxman is not applicable to biologic-based drugs; exclusivity periods are
longer and no mechanism comparable to a Paragraph IV (for chemical-based drugs) exists.
Future work should explore whether the rotation documented by (Branstetter et al., 2011)
_ENREF_8into biologics is also occurring with respect to externally acquired patents. Ultimately,
we need to understand how this rotation and changing nature of innovation impacts overall
welfare and the balance between access and innovation that Hatch-Waxman tried so hard to
create. As with all research much remains to be done.
31
Allison JR, Lemley MA. 1998. Empirical Evidence on the Validity of Litigated Patents. American Intellectual Property Law Association (AIPLA) Quarterly Journal 26.
Angrist JD, Pischke J-S. 2008. Mostly harmless econometrics: An empiricist's companion. Princeton university press.
Arora A, Ceccagnoli M. 2006. Patent Protection, Complementary Assets, and Firms’ Incentives for Technology Licensing. Management Science 52(2): 293-308.
Arora A, Fosfuri A, Gambardella A. 2001. Markets for Technology: The Economics of Innovation and Corporate Strategy. MIT Press: Cambridge, MA.
Arora A, Gambardella A. 2010. Ideas for rent: an overview of markets for technology. Industrial and Corporate Change 19(3): 775-803.
Arora A, Gambardella A, Magazzini L, Pammolli F. 2009. A Breath of Fresh Air? Firm Type, Scale, Scope, and Selection Effects in Drug Development. Management Science 55(10): 1638-1653.
Berndt E, Mortimer R, Parece A. 2007. Do Authorized generic drugs deter paragraph IV certifications? Recent evidence. In Working Paper.
Branstetter LG, Chatterjee C, Higgins M. 2011. Regulation and Welfare: Evidence from Paragraph IV Generic Entry in the Pharmaceutical Industry. National Bureau of Economic Research Working Paper Series No. 17188.
Branstetter LG, Chatterjee C, Higgins M. 2014. Starving (or Fattening) the Golden Goose?: Generic Entry and the Incentives for Early-Stage Pharmaceutical Innovation. National Bureau of Economic Research Working Paper Series No. 20532.
Bulow J. 2004. The Gaming of Pharmaceutical Patents. Innovation Policy and the Economy 4(ArticleType: research-article / Full publication date: 2004 / Copyright © 2004 The University of Chicago Press): 145-187.
Cassiman B, Veugelers R. 2006. In Search of Complementarity in Innovation Strategy: Internal R&D and External Knowledge Acquisition. Management Science 52(1): 68-82.
Caves RE, Whinston MD, Hurwitz MA, Pakes A, Temin P. 1991. Patent Expiration, Entry, and Competition in the U.S. Pharmaceutical Industry. Brookings Papers on Economic Activity. Microeconomics 1991(ArticleType: research-article / Full publication date: 1991 / Copyright © 1991 The Brookings Institution): 1-66.
Ceccagnoli M, Graham SJH, Higgins MJ, Lee J. 2010. Productivity and the role of complementary assets in firms demand for technology innovations. Industrial and Corporate Change 19(3): 839-869.
Ceccagnoli M, Higgins MJ, Palermo V. 2014. Behind the Scenes: Sources of Complementarity in R&D. Journal of Economics & Management Strategy 23(1): 125-148.
Ceccagnoli M, Jiang L. 2013. The Cost of Integrating External Technologies: Supply and Demand Drivers of Value Creation in the Markets for Technology. Strategic Management Journal 34(4): 404-425.
Chesbrough H. 2003. Open Innovation: The New Imperative for Creating and Profiting from Technology Harvard Business School Publishing: Boston, MA.
DiMasi JA, Hansen RW, Grabowski HG. 2003. The price of innovation: new estimates of drug development costs. Journal of Health Economics 22(2): 151-185.
Gallini NT. 1984. Deterrence by Market Sharing: A Strategic Incentive for Licensing. American Economic Review 74: 931-941.
Gallini NT, Winter SG. 1985. Licensing in the Theory of Innovation. RAND Journal of Economics 16: 237-252.
Gambardella A, Giuri P, Luzzi A. 2007. The market for patents in Europe. Research Policy 36(8): 1163-1183.
Gans JS, Hsu DH, Stern S. 2002. When Does Start-Up Innovation Spur the Gale of Creative Destruction? The RAND Journal of Economics 33(4): 571-586.
32
Grabowski H. 2004. Are the Economics of Pharmaceutical Research and Development Changing? Productivity, Patents and Political Pressures. PharmacoEconomics 22(Suppl. 2): 15-24.
Grabowski HG, Kyle M. 2007. Generic competition and market exclusivity periods in pharmaceuticals. Managerial and Decision Economics 28(4-5): 491-502.
Hemphill CS. 2006. Paying for delay: pharmaceutical patent settlement as a regulatory design problem. New York University Law Review 81: 1553-1623.
Hemphill CS, Sampat BN. 2011. When Do Generics Challenge Drug Patents? Journal of Empirical Legal Studies, 2011.
Hemphill CS, Sampat BN. 2012. Evergreening, patent challenges, and effective market life in pharmaceuticals. Journal of Health Economics 31(2): 327-339.
Herman MR. 2011. The Stay Dilemma: Examining Brand and Generic Incentives for Delaying the Resolution of Pharmaceutical Patent Litigation. Columbia Law Review 111.
Higgins MJ, Graham SJH. 2009. Balancing Innovation and Access: Patent Challenges Tip the Scales. Science 326(5951): 370-371.
Higgins MJ, Rodriguez D. 2006. The outsourcing of R&D through acquisitions in the pharmaceutical industry. Journal of Financial Economics 80(2): 351-383.
Knowles SM. 2010. Fixing the Legal Framework for Pharmaceutical Research. Science 327: 1083 - 1084. Knowles SM, Higgins M. 2011. Vertical disintegration in the pharmaceutical industry and the role of IP.
Intellectual Asset Management 45: 10-15. Kogut B, Zander U. 1993. Knowledge of the Firm and the Evolutionary Theory of the Multinational
Corporation. Journal of International Business Studies 24(4): 625-645. Lanjouw JO, Schankerman M. 2001. Characteristics of Patent Litigation: A Window on Competition. The
RAND Journal of Economics 32(1): 129-151. Lanjouw JO, Schankerman M. 2004. Patent Quality and Research Productivity: Measuring Innovation
with Multiple Indicators. The Economic Journal 114(495): 441-465. Laursen K, Salter A. 2006. Open for innovation: the role of openness in explaining innovation
performance among U.K. manufacturing firms. Strategic Management Journal 27(2): 131-150. Lemley MA, Shapiro C. 2005. Probabilistic Patents. The Journal of Economic Perspectives 19(2): 75-98. Lerner J, Shane H, Tsai A. 2003. Do equity financing cycles matter? Evidence from biotechnology
alliances. Journal of Financial Economics 67(3): 411-446. McGahan AM, Silverman BS. 2006. Profiting from technological innovation by others: The effect of
competitor patenting on firm value. Research Policy 35(8): 1222-1242. McWilliams A, Siegel D. 1997. Event Studies in Management Research: Theoretical and Empirical Issues.
The Academy of Management Journal 40(3): 626-657. Morgan Stanley. 2010. Pharmaceuticals: exit research and create value. PwC. 2013. 2013 Patent Litigation Study. Big cases make headlines, while
patent cases proliferate. Reiffen D, Ward MR. 2005. Generic Drug Industry Dynamics. Review of Economics and Statistics 87(1):
37-49. Scherer FM. 2010. Pharmaceutical Innovation. In Handbook of the Economics of Innovation. Hall BH,
Rosemberg N (eds.): North Holland. Scott Morton FM. 1999. Entry decisions in the generic pharmaceutical industry. RAND Journal of
Economics 30(3): 421-440. Scott Morton FM. 2000. Barriers to entry, brand advertising, and generic entry in the US pharmaceutical
industry. International Journal of Industrial Organization 18(7): 1085-1104. Shapiro C. 2003. Antitrust Limits to Patent Settlements. RAND Journal of Economics 34(2): 391-411.
33
Somaya D. 2012. Patent Strategy and Management An Integrative Review and Research Agenda. Journal of Management 38(4): 1084-1114.
Teece DJ. 1986. Profiting from technological innovation: Implications for integration, collaboration, licensing and public policy. Research Policy 15(6): 285-305.
Voet MA. 2013. The generic challenge: understanding patents, FDA and pharmaceutical life-cycle management. Universal-Publishers.
Williamson OE. 1985. The economic institutions ol capitalism. Free Press: New York.
Figure.1 Paragraph IV challenge description Source: Re-adapted from Bulow (2004)
Generic manufacturer applies for ANDA and Paragraph IV certification
Pharmaceutical company sues
generic applicant
Court rules in favor of pharmaceutical
company. FDA cannot approve ANDA and
generic manufacturer cannot enter.
Court rules in favor of generic manufacturer.
FDA can approve ANDA and 180 days
exclusivity period is granted to the first generic applicant
Patent expires. FDA can approve ANDA
but 180 days exclusivity period is
not granted
Pharmaceutical company DOES NOT
sue generic applicant
Generic manufacturer may
enter
34
Figure 2. Number of unique drugs challenged per year
Figure 3. Distribution of patents per number of Paragraph IV challenges
6.6
26.4
49.50
20
40
60
Num
ber
of ch
alle
ng
ed
dru
gs
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
417
225
69
32
11 163
0
100
200
300
400
Num
ber
of p
ate
nts
0 2 4 6Number of challenges
35
Table 1. Descriptive statistics
Variable Obs. Mean Std. Dev. Min Max Paragraph IV (patent level) 7013 0.083131 0.2761 0 1 Paragraph IV outcome 676 0.491124 0.500291 0 1 2007 Dummy 7013 0.41751 0.493184 0 1 External Patent 7013 0.499216 0.500035 0 1 Ln(1+Sales) 7013 10.76971 3.099298 0 15.92163 Product Patent 7013 0.25082 0.433516 0 1 Drug Delivery patent 7013 0.136318 0.343151 0 1 Composition patent 7013 0.344218 0.475146 0 1 Method Patent 7013 0.206189 0.404596 0 1 Claims 7013 19.82105 22.02607 1 396 Backward citation 7013 15.49209 32.7743 0 276 Forward citation 7013 9.204477 19.72723 0 428 Newest patent 7013 0.48182 0.499705 0 1 Patent per Innovation 7013 4.079281 3.044918 1 18 Drug importance (Sales) 6805 0.118691 0.234041 0 1 Teva 7013 0.028091 0.165244 0 1
36
Table 2. Correlation Table
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
1.Paragraph IV (patent level) 1
2.Paragraph IV outcome -0.0199 1
3.2007 Dummy 0.1315* -0.1653* 1
4.External Patent -0.0527* -0.0176 0.0007 1
5.Ln(1+Sales) 0.1814* -0.0549 0.1000* -0.1913* 1
6.Product Patent -0.0265* -0.0364 -0.0476* 0.0229 0.0820* 1
7.Drug Delivery patent 0.0143 -0.0233 0.0260* 0.1211* -0.0475* -0.2299* 1
8.Composition patent 0.0003 -0.0056 0.0092 -0.0733* -0.0520* -0.4192* -0.2878* 1
9.Method Patent 0.0380* 0.0511 0.0073 0.0128 0.021 -0.2949* -0.2025* -0.3692* 1
10.Claims 0.0383* -0.0669 0.0451* 0.0482* -0.0733* -0.1173* 0.0869* 0.0232 0.0260* 1
11.Backward citation 0.0484* -0.0282 0.0589* -0.0109 -0.0936* -0.1530* 0.1835* 0.0017 -0.0815* 0.3161* 1
12.Forward citation 0.0200 -0.0222 0.1655* 0.006 0.0852* 0.1013* 0.0106 -0.011 -0.1081* 0.0794* -0.0047 1
13.Newest patent -0.0144 0.033 -0.0664* -0.0239* -0.0555* -0.0688* 0.0186 0.0666* 0.0185 -0.0182 -0.0341* -0.2076* 1
14.Patent per Innovation 0.0685* 0.1621* 0.1362* -0.0445* 0.1799* -0.0864* -0.0484* -0.0660* 0.0663* 0.1128* 0.2000* 0.0668* -0.3887* 1
15.Drug importance (Sales) 0.0804* -0.0126 0.0499* 0.0906* 0.1177* -0.0392* -0.0272* -0.0408* 0.1513* -0.0094 -0.0374* -0.0406* -0.0401* 0.1207* 1
16.Teva 0.4583* -0.0635 0.0748* -0.0334* 0.1252* 0.0191 -0.0223 0.0003 0.0115 0.0062 -0.0233 0.0226 -0.0275* 0.0534* 0.0561* 1
* Indicates that the correlation is significant at the 5%level
37
Table 3. Patent Level: Challenge
Linear Probability Model Probit Logit (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) 2007 Dummy 0.069*** 0.065*** 0.058*** 0.074*** 0.069*** 0.524*** 0.499*** 0.492*** 0.509*** 0.519*** 1.012*** 0.965*** 0.941*** 0.945*** 0.931*** (0.008) (0.008) (0.008) (0.013) (0.012) (0.057) (0.058) (0.055) (0.081) (0.075) (0.115) (0.118) (0.110) (0.159) (0.148) External Patent -0.030*** -0.034*** -0.017** -0.026*** -0.007 -0.221*** -0.245*** -0.030 -0.232** 0.005 -0.437*** -0.486*** -0.053 -0.516*** -0.069
(0.009) (0.009) (0.008) (0.010) (0.009) (0.065) (0.065) (0.059) (0.095) (0.088) (0.126) (0.126) (0.114) (0.198) (0.184) 2007 Dummy* External Patent
-0.018 -0.023 -0.022 -0.061 0.047 0.026 (0.016) (0.016) (0.114) (0.111) (0.231) (0.223)
Product Patent 0.050*** 0.043** 0.049*** 0.043** 0.432** 0.334* 0.432** 0.335* 0.886** 0.739** 0.886** 0.738** (0.019) (0.018) (0.019) (0.018) (0.197) (0.187) (0.198) (0.188) (0.405) (0.376) (0.404) (0.376) Drug Delivery patent
0.063*** 0.063*** 0.063*** 0.063*** 0.512** 0.576*** 0.512** 0.577*** 1.055*** 1.298*** 1.055*** 1.298*** (0.021) (0.020) (0.021) (0.020) (0.201) (0.197) (0.201) (0.197) (0.409) (0.387) (0.409) (0.387)
Composition patent
0.059*** 0.059*** 0.058*** 0.059*** 0.480** 0.508*** 0.480** 0.509*** 0.985** 1.073*** 0.985** 1.073*** (0.019) (0.018) (0.019) (0.018) (0.195) (0.188) (0.195) (0.188) (0.400) (0.373) (0.399) (0.373)
Method Patent 0.082*** 0.077*** 0.082*** 0.077*** 0.661*** 0.616*** 0.661*** 0.618*** 1.319*** 1.287*** 1.318*** 1.287*** (0.020) (0.019) (0.020) (0.019) (0.198) (0.189) (0.198) (0.189) (0.404) (0.375) (0.404) (0.375) Claims 0.000 0.001 0.001 0.001 0.002 0.001 0.002 0.001 0.004 0.002 0.004 0.002 (0.001) (0.001) (0.001) (0.001) (0.002) (0.001) (0.002) (0.001) (0.004) (0.003) (0.003) (0.003) Backward citation
0.001 0.001** 0.001 0.001** 0.002 0.002** 0.002 0.002** 0.003* 0.003*** 0.003* 0.003*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.002) (0.001) (0.002) (0.001)
Forward citation
0.001 0.001 0.001 0.001 0.001 -0.001 0.001 -0.001 0.002 -0.002 0.002 -0.002 (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.003) (0.002) (0.003) (0.002)
Newest patent -0.017* -0.010 -0.017* -0.010 -0.097 -0.017 -0.097 -0.017 -0.177 0.014 -0.177 0.014 (0.009) (0.009) (0.009) (0.009) (0.065) (0.060) (0.065) (0.060) (0.128) (0.114) (0.128) (0.114) Ln(1+Sales) 0.014*** 0.014*** 0.211*** 0.211*** 0.474*** 0.474*** (0.001) (0.001) (0.028) (0.027) (0.045) (0.045) Constant 0.072*** 0.010 -0.146*** 0.006 -0.152*** -1.721*** -2.211*** -4.680*** -2.218*** -4.700*** -3.107*** -4.110*** -9.731*** -4.097*** -9.725*** (0.007) (0.021) (0.022) (0.021) (0.022) (0.061) (0.203) (0.411) (0.207) (0.408) (0.125) (0.417) (0.689) (0.425) (0.687) Obs. 7,013 7,013 7,013 7,013 7,013 7,013 7,013 7,013 7,013 7,013 7,013 7,013 7,013 7,013 7,013 Clustered Standard errors in parentheses * p < 0.11, ** p < 0.05, *** p < 0.01
38
Table 4. Patent Level: Outcome
Linear Probability Model Probit Logit (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) 2007 Dummy -0.165*** -0.178*** -0.166*** -0.239*** -0.226*** -0.560*** -0.600*** -0.579*** -0.854*** -0.834*** -0.949*** -1.023*** -0.991*** -1.464*** -1.434*** (0.041) (0.042) (0.042) (0.052) (0.053) (0.147) (0.153) (0.155) (0.214) (0.217) (0.254) (0.268) (0.272) (0.379) (0.383) External Patent 0.013 0.033 0.037 -0.055 -0.049 0.041 0.090 0.106 -0.230 -0.215 0.070 0.156 0.182 -0.393 -0.370
(0.043) (0.044) (0.045) (0.065) (0.064) (0.147) (0.146) (0.152) (0.221) (0.222) (0.248) (0.248) (0.259) (0.379) (0.382) 2007 Dummy* External Patent
0.145* 0.141* 0.531* 0.539* 0.910* 0.926* (0.084) (0.085) (0.293) (0.299) (0.506) (0.517)
Product Patent -0.081 -0.074 -0.089 -0.083 -0.240 -0.239 -0.278 -0.273 -0.403 -0.400 -0.471 -0.463 (0.134) (0.132) (0.138) (0.136) (0.462) (0.468) (0.487) (0.489) (0.779) (0.790) (0.825) (0.830) Drug Delivery patent
-0.040 -0.042 -0.042 -0.045 -0.098 -0.115 -0.107 -0.125 -0.165 -0.192 -0.180 -0.210 (0.141) (0.139) (0.145) (0.142) (0.479) (0.486) (0.502) (0.507) (0.807) (0.822) (0.849) (0.860)
Composition patent
-0.038 -0.035 -0.048 -0.045 -0.103 -0.106 -0.143 -0.144 -0.169 -0.175 -0.238 -0.239 (0.132) (0.129) (0.136) (0.134) (0.454) (0.459) (0.479) (0.481) (0.764) (0.774) (0.810) (0.815)
Method Patent -0.066 -0.048 -0.077 -0.059 -0.165 -0.128 -0.214 -0.174 -0.274 -0.210 -0.357 -0.286 (0.133) (0.130) (0.137) (0.135) (0.461) (0.465) (0.486) (0.488) (0.776) (0.785) (0.823) (0.827) Claims -0.001* -0.001** -0.001** -0.001** -0.004* -0.005* -0.005* -0.005** -0.007* -0.008* -0.008* -0.009* (0.001) (0.001) (0.001) (0.001) (0.003) (0.003) (0.003) (0.003) (0.004) (0.004) (0.004) (0.005) Backward citation
0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.002 0.002 0.002 0.002 (0.001) (0.001) (0.001) (0.001) (0.001) (0.002) (0.001) (0.002) (0.002) (0.003) (0.002) (0.003)
Forward citation 0.001 0.001 0.001 0.001 0.001 0.002 0.001 0.001 0.002 0.003 0.002 0.003 (0.001) (0.001) (0.001) (0.001) (0.003) (0.003) (0.003) (0.003) (0.005) (0.006) (0.005) (0.006) Newest patent 0.074 0.076 0.072 0.075 0.267* 0.277* 0.267* 0.279* 0.448* 0.466* 0.449* 0.471* (0.047) (0.048) (0.047) (0.048) (0.159) (0.166) (0.164) (0.170) (0.270) (0.281) (0.279) (0.289) Patents per Innovation
0.028*** 0.030*** 0.027*** 0.030*** 0.099*** 0.108*** 0.100*** 0.110*** 0.167*** 0.184*** 0.170*** 0.188*** (0.007) (0.008) (0.008) (0.008) (0.027) (0.029) (0.028) (0.030) (0.047) (0.050) (0.049) (0.052)
Teva 0.042 0.057 0.041 0.056 0.015 0.079 0.030 0.091 0.055 0.165 0.084 0.190 (0.045) (0.045) (0.045) (0.045) (0.173) (0.181) (0.179) (0.186) (0.300) (0.316) (0.312) (0.327) Drug importance (Sales)
-0.187* -0.185* -0.622* -0.647* -1.071* -1.114* (0.101) (0.099) (0.356) (0.358) (0.618) (0.623)
Constant 0.587*** 0.478*** 0.486*** 0.530*** 0.535*** 0.303** -0.080 -0.046 0.110 0.142 0.512** -0.145 -0.088 0.180 0.236 (0.038) (0.139) (0.137) (0.147) (0.145) (0.136) (0.477) (0.485) (0.514) (0.519) (0.233) (0.802) (0.816) (0.870) (0.878) Obs. 676 676 662 676 662 676 676 662 676 662 676 676 662 676 662 Clustered standard errors in parentheses * p < 0.11, ** p < 0.05, *** p < 0.01
39
Table 5. Event study results divided by subsample
Pharmaceutical wins subsample Time Window (days) N CAR Z-statistics
(-1, +1) 75 -0.01% -0.818 (-3, +1) 75 -0.02% -0.818 (-5, +1) 75 -0.09% -0.125
Pharmaceutical loses subsample Time Window (days) N CAR Z-statistics
(-1, +1) 69 -0.53% 0.487 (-3, +1) 69 -0.82% -1.519* (-5, +1) 69 -0.83% -2.001**
The symbols *, **, and *** denote statistical significance at the 0.10, 0.05, 0.01 levels, respectively. Time windows are with respect to the event date set as t=0